The questions that actually predict problems

The standard list of questions to ask a potential roommate is mostly a personality quiz. Are you a morning person. How tidy are you, one to ten. Do you like having people over. Everyone answers these the same way — moderately, seven, sometimes — and nobody learns anything.

The problem isn’t that the questions are stupid. It’s that they ask people to self-report on traits where self-perception is unreliable and mildly flattering. Nobody rates themselves a three on tidiness. The person who leaves pans in the sink for four days rates themselves a seven, sincerely, because they mean it about their own room.

Questions that work do one of three things: ask about the past instead of the disposition, ask about logistics that have factual answers, or ask something where an evasive answer is itself informative.

Before anything: what isn’t a screening question

If you’re the one choosing — you have the spare room, they’re applying — selection and advertising law may apply to you, and it varies by jurisdiction. Independent of that: race, religion, national origin, disability, family status, sex and sexual orientation are not screening criteria, here or anywhere on this site. Everything below is about schedules, money, habits and history.

That’s not a constraint on the useful questions. It rules out none of them.

Ask about the past, not the disposition

“How tidy are you” invites self-assessment. “What’s the kitchen situation in your current place” invites description, and description leaks.

Some that work:

“Why are you moving?” The most productive question available and almost nobody asks it directly. The answers are usually mundane — job, lease ending, partner, city. Occasionally they’re a story about the last household, and the way somebody narrates a conflict tells you a great deal about how the next one will go. Note whether anything was their fault in their own telling.

“How long were you at the last place?” A pattern of four-month stays is a real signal. One short stay is a life event; four in a row is a trajectory.

“What’s the current place like?” Open enough that what they choose to describe is the data. Some people describe the flat. Some describe the people. Some describe the commute. Some describe what’s wrong with it.

“When did you last have a problem with a housemate, and how did it get resolved?” Slightly confrontational and worth it. “Never” is not a great answer — everyone has, and it usually means either avoidance or a short history. The good answer is a small specific thing that got talked about.

Ask logistics, which have facts underneath

These are the questions with checkable answers, and where the answers matter more than anyone expects.

Working hours and days. Not “are you a morning person” but “what time do you leave and get back”. Two people on the same schedule share a bathroom queue and a kitchen rush. Two people on opposite schedules share nothing and one of them sleeps while the other is awake in a flat with thin walls. Both are workable; both are worth knowing in advance.

Work from home, and how much. This has become the single biggest determinant of what a shared flat is actually like, and it’s often not discussed. Someone home five days a week with calls is a different housemate from someone in an office five days a week, in ways that cover heating, kitchen use, noise, and the question of whether the living room is a workspace.

Partners. Not “do you have a partner” but “would they be around much”. The near-universal houseshare grievance is the partner who effectively lives there without contributing, and it is close to impossible to renegotiate once established.

Who’s on the tenancy and who pays whom. A factual question with a factual answer. What the various arrangements are called and what they entail is law and varies by jurisdiction — but you should know the shape of it before you decide, and a local tenant service can explain what that shape means where you are.

Bills: included or not, and what they’ve been running. Ask for the actual recent figure, not an estimate. People estimate low.

What happened to the last person in the room, and when did they leave? Combines a factual answer with a soft spot. “They moved in with their partner, three weeks ago” is a complete answer. Vagueness here is one of the more reliable tells in the whole process.

Questions where evasion is the answer

A small category, and the most efficient of the three, because you learn from the response either way.

“Can I meet the other housemates?” In a shared flat, the right answer is some version of yes — they might be out, they might be at work, but the request should land as normal. A reason why not is worth listening to carefully.

“How many people live here in total?” Should be instant. Hesitation, or an answer that changes during the viewing, means someone isn’t counted — a partner, a sublet, someone in the living room.

“Is there anything about the flat you’d want to know if you were me?” Some people say no. Some people tell you about the boiler, the neighbour, or the housemate who plays bass. The ones who tell you are the ones you want.

Flag check

FLAG CHECK — the screening conversation

  · Vague about why the last person left
                    → one of the strongest signals
                      available. Press once, politely.

  · Won't let you meet existing housemates
                    → WALK AWAY, in a shared flat.
                      There's a reason.

  · Headcount answer shifts during the viewing
                    → WALK AWAY. Someone is uncounted
                      and it will be your problem.

  · Every past conflict was the other person's fault
                    → pattern worth weighting. Not
                      disqualifying on its own.

  · Rates themselves highly tidy
                    → NOT A FLAG. Everyone does.
                      Carries no information.

  · Unusual working hours
                    → NOT A FLAG. A difference. Only
                      matters against your own hours
                      and the wall thickness.

  · Nervous, awkward, over-explaining
                    → NOT A FLAG. This is a stressful
                      conversation for both of you.

Asking without conducting an interview

The failure mode of a good question list is delivering it as one. Somebody reading from their phone while you stand in a hallway does not produce candid answers; it produces the answers people give in interviews.

Most of this fits into ordinary conversation during a viewing. Why are you moving comes up by itself. Working hours arrive via the commute. What happened to the last person arrives via what’s still in the room. Four or five deliberate questions is plenty, and the rest is noticing.

If you’re the one with the room, the same applies in reverse and more strongly: the person you’d most want is also assessing you, and a checklist interrogation is a reason to choose the other place.

What twenty minutes can’t reach

All of this is self-report from someone with an incentive, filtered through a short conversation in which both of you are performing slightly.

You will not find out that someone is passive-aggressive about the heating, or that they have a friend who stays for a fortnight twice a year, or that they are lovely and impossible. Those things surface in month two.

What the questions do is catch the knowable problems — the uncounted resident, the four-month pattern, the partner who’s really a tenant, the room that three people have left this year. That’s a real subset, and it’s the subset you can actually do something about while you still have a choice.